Optimal Heart Team Protocol to Improve Revascularization Decisions in Patients with Complex Coronary Artery Disease: A Sequential Mixed Method Study

Abstract

Aims

Current guidelines recommend a heart team in the decision-making for patients with complex coronary artery disease (CAD). However, the decision-making stability of these teams has not been evaluated and the optimum protocol is unknown. We assessed inter-team agreement for revascularization decision-making and influencing factors to inform the development of a heart team protocol.

Methods and results

This sequential, explanatory mixed methods study included (i) a cross-sectional quantitative study to assess inter-team agreement on treatment strategy for retrospectively enrolled complex CAD patients and (ii) a qualitative study that used semi-structured interviews with heart team members to identify factors influencing decision-making discrepancy. We randomly selected 101 complex CAD patients. Sixteen specialists were randomly assigned to four heart teams to make decisions for these patients. The primary outcome kappa of inter-team decision-making agreement was moderate (kappa 0.58). Factors influencing decision-making were generated through inductive thematic analysis and were summarized by 3 themes (specialist quality, team composition, and meeting process) and 10 subthemes. Recommendations of heart team implementation were generated based on qualitative and quantitative data at five levels: specialist selection, specialist training, team composition, team training, and meeting process. A detailed protocol on the integration of guidelines, previous experience, and recommendations was generated to establish and deploy a qualified heart team.

Conclusion

Agreement between heart teams for revascularization decision-making in complex CAD patients was moderate. Potential factors associated with decision discrepancies were summarized and recommendations were generated. A detailed heart team protocol was designed and should be validated in future.

Introduction

Decision-making based on the multidisciplinary heart team approach has been shown to improve the appropriateness of revascularization decisions and contribute to better clinical outcomes.^1–3 To optimize the treatment strategies for patients with complex coronary artery disease (CAD), clinical guidelines recommend a heart team consisting of clinical/non-invasive cardiologists, interventional cardiologists, and cardiac surgeons.^4,5 However, detailed recommendations for heart team establishment and implementation are not provided in the current guidelines, potentially limiting the utilization of the heart team approach in clinical practice and leading to reduced care quality.

Prior studies have noted issues with inter-specialist decision-making variability and intra-team non-reproducibility, which affect the stability of heart team decision-making. For example, Denvir et al. found poor agreement in decision-making between different specialists when clinicians were required to make decisions independently (kappa 0.26).⁶ In addition, several studies found that in rediscussions 9–12 months later, nearly 20% of the decisions on revascularization by the heart team changed.^7,8 There is limited evidence, however, on the agreement between heart teams and factors potentially influencing decision-making agreement.

This sequential explanatory mixed methods study aimed to evaluate the inter-team agreement in revascularization decision-making for stable patients with complex CAD. In addition, we assessed potentially influential factors for inter-team agreement and provided recommendations and a heart team protocol to improve decision agreement.

Methods

Design overview

This sequential explanatory mixed methods study consisted of (i) a single-centre, cross-sectional study to assess the agreement in heart team decision-making and (ii) a qualitative investigation to identify the factors potentially influencing agreement (Supplementary material online, Figure S1). All participating specialists provided written informed consent for the study and interview. This study was approved by the Institutional Review Board of the National Center for Cardiovascular Diseases (Beijing, China), and it is registered at ClinicalTrials.gov (NCT04217031).

Cross-sectional study on agreement

Patients with stable CAD according to the National Cardiovascular Data Registry (NCDR) CathPCI criteria (stable angina, no or silent myocardial ischaemia) and angiographically confirmed three-vessel disease or left main disease were eligible for inclusion in the study. The exclusion criteria included (i) prior percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG); (ii) cardiac troponin I (CTnI) greater than the local laboratory upper limit of normal or recent myocardial infarction with CTnI levels still elevated; (iii) concomitant severe valvular disease, macrovascular disease, or huge ventricular aneurysm requiring surgery; or (iv) concomitant atrial fibrillation or severe arrhythmia. Eligible cases were randomly selected from a prospective registry of consecutive patients who underwent coronary angiography between August 2016 and August 2017.⁹

The inclusion criteria for the heart team specialists are shown in the Supplementary material online, Table S1. Eligible specialists were randomly enrolled and assigned to the four heart teams. There were two 3-member heart teams (heart teams 1 and 2) comprising 1 interventional cardiologist, 1 non-interventional cardiologist, and 1 cardiac surgeon and two 5-member heart teams (heart teams 3 and 4) comprising 2 interventional cardiologists, 1 non-interventional cardiologist, and 2 cardiac surgeons.

Heart team meetings were conducted according to the procedure widely used in previous studies.^10–12 All clinical information, coronary angiography, and other available imaging modalities (e.g. electrocardiograph and echocardiography) on included cases were inputted into an electronic meeting support system by a non-clinical coordinator. This information included (i) patients’ demographic and clinical characteristics extracted from medical records according to the NCDR CathPCI data definitions¹³; (ii) clinical presentation, Canadian Cardiovascular Society class, and anti-ischaemic therapy history obtained via interviews by well-trained research nurses during the index hospitalization; and (iii) SYNTAX (Synergy Between Percutaneous Coronary Intervention With Taxus and Cardiac Surgery) score, SYNTAX II score (www.syntaxscore.com), Society of Thoracic Surgeons (STS) score (riskcalc.sts.org), and the European System for Cardiac Operative Risk Evaluation (EuroSCORE) I/II (www.euroscore.org) evaluated by an angiographic core laboratory using a computer-based automatic calculator. Specialists in the four heart teams first made their decisions independently in the electronic system. All specialists were required to choose from among five treatment categories (PCI, CABG, PCI/CABG equipoise, medical therapy, or further testing) before (round I) and after (round II) the heart team discussion. The final treatment strategy was determined by majority decision. All participating heart team members were invited to complete the Ten-Item Personality Inventory (TIPI) in Chinese (Supplementary material online, Table S2), which has been shown to have good psychometric properties.¹⁴

Qualitative investigation on influencing factors

Standard in-depth semi-structured interviews were conducted with all participating specialists during November 2020 to January 2021. The interviews included open-ended questions and subsequent probing questions (Supplementary material online, Table S3), which were chosen based on previous multidisciplinary team frameworks and observations from the meeting discussions.¹⁵ Two experienced qualitative investigators with medical backgrounds conducted these interviews in a quiet and private space where the interviewees could share their views freely and confidentially without outside influence. One researcher conducted the interview, while the other took notes and prompted the discussion. All interviews were audiotaped and professionally transcribed.

Data analysis

Data are presented as means ± standard deviations for continuous variables and percentages for discrete variables. In the quantitative analysis to assess decision-making agreement, Fleiss's (more than two raters) and Cohen's (two raters) kappa coefficients were used to evaluate inter-team, inter-specialist, and inter-round agreement for treatment decisions. Mean decision time was also calculated. Analyses were conducted using SPSS version 26 (IBM Inc., Chicago, IL, USA).

In the qualitative analysis to identify influencing factors, qualitative data were analysed and reported according to the Consolidated Criteria for Reporting Qualitative Research (COREQ) 32.¹⁶ An inductive thematic analysis was applied to generate themes and subthemes related to inter-team discrepancy. Two analysts (S.L. and H.M.) coded each transcript independently and met to discuss the generation of themes and coding guidelines. All coding discrepancies were discussed and resolved by negotiated consensus. Quotes were translated into English and back-translated to Chinese. The back-translation was compared with the original text to verify the accuracy of the translation. After the themes and subthemes were established, we inferred how each subtheme affected decision-making by constantly comparing and integrating both qualitative and quantitative data. A visual joint display was used to present these results.¹⁷ Analyses were performed with Nvivo V12.0 Plus qualitative software (QSR International Version 12, 2018).

The sample size for the quantitative study was calculated based on a hypothesized almost-perfect decision-making agreement (kappa 0.80)¹⁸ (Supplementary material online, Table S4). We defined the lower bound of the 95% two-sided confidence interval for the kappa value as 0.60, and the power was 0.80. Based on a previous study, we assumed the following distribution of treatment decisions: 26.0% CABG, 17.0% PCI or equipoise, and 57.0% further testing or medical therapy.⁸ Given these assumptions, a sample size of 101 patients was calculated for the study.¹⁹ For the qualitative study sample size, the concepts of information power and thematic saturation were used for purposive sampling.²⁰ Interviews continued until there were no new concepts emerging from the data; this was achieved at 12 interviews.

Results

Characteristics of cases and specialists

In the final study, 101 patients were included (Supplementary material online, Figure S2) and baseline characteristics are presented in Table 1. The mean age was 60.3 ± 10.8 years, and 78 (77.2%) patients were male. The proportion of patients with left ventricular ejection fraction ≤40% was 6.9%. With respect to coronary anatomic complexity, 19 (18.8%) patients had left main and 97 (96.0%) had three-vessel CAD. The mean SYNTAX score was 21.0 ± 11.1, and the median STS score mortality rate was 0.51%.

Among the 48 specialists who met our eligibility criteria, 16 were randomly selected and enrolled, including 6 cardiac surgeons, 6 interventional cardiologists, and 4 non-interventional cardiologists (Supplementary material online, Figure S3). Baseline characteristics of the enrolled specialists are presented in Table 2. The median practice duration was 23.0 (interquartile range, 19.0–27.0) years, and 10 (62.5%) specialists were male. Personality varied by specialist discipline.

Heart team decision-making agreement

The heart team decision and inter-team agreement are listed in Table 3 and the Supplementary material online, Table S5. Overall inter-team agreement for the three-category decision (CABG, PCI/equipoise, and further testing/medical therapy) was moderate (kappa 0.58). Inter-team agreement ranged from moderate to substantial (kappa 0.51–0.63) when the pairwise comparison was used and was better for the two 5-member teams (team 3/4, kappa 0.60) compared with the two 3-member teams (team 1/2, kappa 0.54) (Supplementary material online, Table S5). Inter-team agreement further decreased (kappa 0.47) when using the five-category decision (CABG, PCI, equipoise, further testing, and medical therapy). Inter-team agreement was slight (kappa 0.16) for equipoise and fair (kappa 0.25) for further testing. In the patient subgroup analysis, inter-team agreement decreased for patients with age <70 years, ejection fraction <50%, body mass index <19.0 kg/m² or >24.0 kg/m², SYNTAX score indicating moderate risk, severe calcification, and severe tortuosity (Supplementary material online, Table S6).

Inter-specialist agreement was better in round II (kappa 0.72–0.85) compared with round I (kappa 0.35–0.55). Within the disciplines, inter-specialist agreement in round I decisions was moderate among interventional cardiologists (kappa 0.56) but fair among cardiac surgeons (kappa 0.42) and non-interventional cardiologists (kappa 0.43). Inter-round agreement was also better in interventional cardiologists (kappa 0.83) compared with cardiac surgeons (kappa 0.73) and non-interventional cardiologists (kappa 0.69).

Factors potentially influencing decision-making agreement

A total of 3 themes and 10 subthemes related to decision-making were identified and summarized (Table 4 and Supplementary material online, Figure S4). The three themes were (i) specialist quality, (ii) team composition, and (iii) meeting process. The 10 subthemes were (i) decision thought process, (ii) understanding of illness and evidence, (iii) understanding of other disciplines, (iv) personality, (v) learning curve, (vi) technical and administrative positions, (vii) number of team members, (viii) discipline selection, (ix) ratio of different disciplines, and (x) meeting process.

Recommendations for improving decision-making agreement

After constantly comparing and integrating quantitative and qualitative findings, we identified how each subtheme affected decision-making (Table 4). Based on these results, the following five recommendations were formulated and are reported in Table 5. (i) specialist selection should not only ensure a base level of technical knowledge/experience for each team member but also consider the balance of technical skill and administration position and personalities (surgeons and non-interventional cardiologists only) among members. (ii) Specialist training should include a pre-meeting discussion to achieve consensus on the decision thought process, the understanding of key decision variables, and evidence and advanced developments in PCI and CABG, especially PCI. (iii) The optimal team composition for routine heart team meetings does not require a non-interventional cardiologist. In addition, there may be more decision stability with 2 surgeons and 2 interventional cardiologists than with 1 surgeon and 1 interventional cardiologist. (iv) Team training should include a pilot discussion of 25–50 cases to ensure team members are comfortable with each other and the decision process. (v) For the meeting process, any meeting form works well, and an online format is encouraged.

Discussion

In this sequential explanatory mixed methods study, we found moderate agreement between heart teams in revascularization decision-making for patients with complex CAD. Subsequent qualitative results identified several factors potentially influencing decision discrepancy. Based on both the quantitative and qualitative findings, we identified how each factor might affect decision-making and provided five recommendations to improve decision-making agreement. We integrated our recommendations, previous experience, and clinical guidelines to develop a protocol for establishing and implementing the heart team (Figure 1 and Supplementary material online, Figure S5).

Previous studies have assessed intra-team non-reproducibility and shared the basic principles of heart team organization and implementation.^6–8,^10–12 However, there has been limited evidence on the stability of inter-team decision-making and little guidance on how best to establish a qualified heart team. This sequential explanatory mixed methods study was conducted to answer these questions. Quantitative efforts showed only moderate agreement (kappa 0.58) in revascularization decisions between heart teams. Both quantitative and qualitative efforts provided an evidence-based heart team protocol that is detailed enough for any institution to establish and implement a qualified heart team. The protocol was derived from the guidelines, our recommendations, and previous experience. The recommendations regarding the volume–outcome relationship in current guidelines were used to establish specialist inclusion criteria,^4,21 and that based on the quantitative and qualitative study results informed the specialist selection/training and team composition/training. Previous experience from other institutions informed the meeting process component of the protocol.

The utility of our findings and protocol are supported by the rigorous design and execution of our study. First, we used an explanatory sequential design to facilitate better understanding of reported discrepancies in heart team decision-making.²² In the quantitative phase of the study, in addition to assessing decision-making agreement, we observed the entire heart team discussion to help inform interview sampling and development of the interview outline for the qualitative phase. In the qualitative phase, we conducted standardized in-depth interviews and conducted qualitative analysis to identify themes for the influencing factors. We also integrated quantitative and qualitative results to further confirm our findings. Second, our sample sizes were fully considered and suitable to justify our study conclusions. For the quantitative analysis, patient sample size was calculated in advance to ensure adequate power to assess decision-making agreement. The sample size for the qualitative interviews was based on theoretical saturation and was achieved after 12 interviews. Third, our study followed guidelines for mixed methods research in healthcare.¹⁶,^23–25 For qualitative data collection, we used semi-structured interviews as well as employing member checks, audiotaping, professional transcription, and back-translation. In addition, multiple experienced researchers of different backgrounds coded and analysed interview data using grounded theory. Taken together, these steps helped ensure the reliability and scientific basis of our findings.

Contrary to current revascularization guidelines, our results did not support the necessity of non-interventional cardiologists in routine heart team meetings. Guidelines and reports on institutional experiences with the heart team approach recommend a heart team that includes clinical/non-interventional cardiologists, interventional cardiologists, and cardiac surgeons.⁴,^10–12,²⁶ Non-interventional cardiologists have been considered a necessary component of the heart team to minimize specialty bias in decision-making because they have the requisite knowledge regarding cardiovascular procedures and represent a neutral party with no conflict of interest.^4,27 However, in this study, both qualitative and quantitative data demonstrated that non-interventional cardiologists were an unstable factor in revascularization decision-making. This may be attributable to a lack of training in PCI and CABG procedures, potentially making it difficult for non-interventional cardiologists to accurately assess the benefits and risks of procedures.²⁷ In addition, evidence from a recent single-centre study demonstrated the effectiveness of a heart team without a non-interventional cardiologist for revascularization decision-making.^26,28 Thus, we recommend the inclusion of non-interventional cardiologists only if necessary.

Adjustments may be needed when using our protocol in real-world clinical practice. In terms of team composition, we found that more members may lead to better inter-team agreement, and a 4-person team including 2 interventional cardiologists and 2 cardiac surgeons was preferred over a 2-person team including a member from each discipline. For hospitals with few qualified specialists, a 2-person team with 1 interventional cardiologist and 1 cardiac surgeon could be used; however, in this circumstance, even greater care should be taken to follow the other recommendations in the protocol. In terms of specialist training, clinicians’ busy schedule could pose a challenge to the pre-meeting conferences. However, due to the advances in the field of communication technology, other alternatives for conducting pre-meetings could be considered, such as online video conferences or virtual reality techniques.^29,30 Our goal in the current study was to reduce the decision-making bias resulting from team establishment and implementation, including specialist selection/training, team composition/training, and the meeting process. To further standardize heart team practice and improve decision-making, our protocol should be combined with other multidisciplinary team principles such as effective patient involvement and patient-centred decision-making.^11,15

Study limitations

Our study has several limitations. First, patients and specialists were enrolled from a single centre, potentially limiting the generalizability of the results to other institutions. To reduce this bias, we randomly selected patients and specialists from a large centre with diverse patient and specialist populations. Second, because we used past cases for the heart team discussions, the specialists in our study were unable to have face-to-face consultations with patients and, therefore, did not obtain a clinical history or perform a physical examination first-hand. It is possible that clinician decisions are influenced by direct contact with the patient. Third, given the potential biases of qualitative analysis, the themes and related recommendations identified in our study should be considered hypotheses. A randomized controlled trial is needed to validate the heart team protocol.

Conclusions

The level of inter-team agreement for revascularization decisions in patients with complex CAD was moderate. Potential factors associated with decision discrepancies were summarized by three themes, i.e. specialist quality, team composition, and meeting process. Based on both quantitative and qualitative results, we identified how each influencing factor contributed to decision-making and provided five recommendations at five levels for the implementation of the heart team approach, i.e. (i) specialist selection, (ii) specialist training, (iii) team composition, (iv) team training, and (v) meeting process. A detailed heart team protocol based on these recommendations and previous evidence was generated to establish and deploy a qualified heart team.

Acknowledgements

The authors thank Drs Xianqiang Wang, Yan Yang, Ge Gao, Jie Qian, Lei Song, Dong Yin, Jie Zhao, Jun Zhang, Weihua Song, Rong Liu, and Bo Xu for their assistance with the heart team and collection of heart team implementation data.

Funding

This work was supported by the National Key Research and Development programme (2016YFC1302000), Beijing Municipal Commission of Science and Technology project (D171100002917001), and Basic Research Funds for the Central Universities project (2019-XHQN14). The funding sources were not involved in study design, data collection, analysis, interpretation of data, writing the report, and the decision to submit the article for publication.

Conflict of interest

None declared.

Data availability

The data underlying this article cannot be shared publicly because participants in the study did not provide consent for their data to be made widely available. However, the authors will consider reasonable requests for access to summary information.

References

Author notes